Aloidendron dichotomum

Subpopulations are scattered into discrete subpopulations, some with >100 km between them, and range in size from <100 to 50,000+ individuals. Observations have been largely restricted to road-sides and since roads are sparse in this vast arid region, much of the range remains unsurveyed. Aloidendendron dichotomum occurs in both summer and winter rainfall areas, varying topographies and on multiple soil types and gradients (Foden et al. 2007). The reasons for the discontinuities between subpopulations are unknown.

Because dead trees decay very slowly in the species’ arid habitat, often remaining standing for at least a decade, it is possible to measure each subpopulation’s relative mortality. Similarly, individuals’ slow growth (they reach maximum height at approximately 80–120 years) and clear age stages to senescence, at least 200 years (Vogel 1974) and possibly up to 350 years (Foden et al. 2007), make subpopulation age structure and recruitment history apparent.

Ongoing long-term monitoring of 40 subpopulations from across the species’ range spans 18 years to date (2003 to 2021), with all surveyed in 2003, most again in 2008, and a subsample of sites resurveyed in 2018, 2020 or 2021 (Foden et al., unpublished). The most northerly subpopulations remain to be surveyed after 2008. Calculated using a 100-year ‘shifting window’ from 2003–2102, based on the longest of the monitoring intervals measured since 2003, an overall population decline of 26.2% by 2102 was inferred, assuming linear population decline rates. Further, by 2102 sixteen individual subpopulations are projected to become extinct, nine to decrease by >30% and seven to increase by >30%. In eight subpopulations no juveniles were found at any point during the survey. Twenty-four subpopulations are projected to lose all juveniles by 2102, but six to increase juvenile percentage by >30%. With higher levels of mortality present in northerly populations once the northern subpopulations are resampled it is likely that the population decline projections to 2102 will exceed 30%.

A climate change explanation for A. dichotomum's suspected decline is supported by three independent lines of evidence. Firstly, observed declines correspond to changes in the region’s meteorological records (Foden et al. 2007, IPCC 2021) which show marked increases in mean temperatures and decreases in water balance during the monitoring period. These may have led to surpassing of the species’ physiological thresholds in subpopulations in the warmest and driest parts of its range.

Secondly, based on observations (Foden 2002, Foden et al. 2007, Jack et al. 2016) and long-monitoring (van der Merwe and Geldenhuys 2017, Foden et al. unpublished), clear patterns of mortality and recruitment occur across latitude and elevation. The greater mortality in northern (i.e. equatorial, warm range-edge) and lower elevation subpopulations, and population age structure with a relatively high percentage of juveniles in southern populations is a characteristic signal of climate change impacts on a sedentary species (Foden et al. 2007).

Thirdly, species distribution models for A. dichotomum project losses of between 33% and 68% of currently suitable habitat space by 2070 based on RCPs 2.6 (likely emission scenario) and 8.5 (worst case scenario) respectively (Hausfather and Peters 2020). Models predict habitat gains which range from 84% and 99% by 2070, also based on RCPs 2.6 and 8.5 respectively (see attached Supporting Information document). The spatial patterns of change in habitat suitability approximately follow theory-based expectations for climate change, with losses more prevalent in warmer and drier, more northerly subpopulations as physiological tolerances of heat and water availability are crossed (Grey 2019). Modelled gains occur in more southerly range areas, which were historically too cold for the species, and areas of no change to 2070 remain in the southeast and along a mid-longitude mountainous ‘spine’ running north–south through much of the latitude range, under both RCPs. In order to predict if dispersal into modelled future suitable habitat is possible we considered palaeo-range reconstruction which indicates that this species would have retreated poleward and westward to climate refugia during the Last Glacial Maximum (LGM) where conditions would have remained suitable for the species to persist (Brodie et al. 2021). During this time, models predict that suitable habitat space for the species would have declined by 69% relative to its current distribution (Brodie et al. 2021). Following this, the species would have expanded its geographic range to its current extent at a rate of ~0.4 km/decade (Brodie et al. 2021). Subsequent modelling of its future geographic range predict a range expansion of the species by 191 km eastwards, with suitable habitat space being lost in the most northerly and westerly extents of its current range (Brodie et al. 2021). This range shift would require the species to migrate at a rate of 6 km/decade, which is a rate 15 times faster than its historical migration rate (Brodie et al. 2021), and one that is unlikely to be possible for such a slow-growing species. Since the species has not extended its range to date, despite clear climate change impacts, and the fact that there is high pressure on juvenile recruitment from wildlife and livestock grazing, and ongoing habitat degradation, a zero dispersal scenario is assumed. Under likely emission scenarios the population is predicted to decline by 33% by 2070 and is suspected to exceed 35% by the year 2102.